Collection of self published OpenShift 4 tools
Intended use is in a cron or other secheduled jobs to backup a cluster etcd via
ocp4-backup.sh clustername
The clustername is used as a predefined list of clusters that provide 4 pieces of data
- ETCD_HOST - What server to ssh to and preform the snapshot
- ETCD_USER - The username (core in most cases) to ssh as
- ETCD_KEYFILE - The ssh keyfile for the user
- LOCAL_BACKUP_DIR - Full path of where to store the file
- BACKUP_FILE - A way to customize the backup file name
Replicating the conditional test in the case like nonprod will provide a way to introduce multiple clusternames.
If you need to change your NTP resources from generic internet sources you can use custom machine configs.
| Filename | Usage |
|---|---|
| 99-ntp-sources-master.yaml | Master config |
| 99-ntp-sources-worker.yaml | Worker config |
Each of these files use label selectors in the metadata to apply to either the master or worker nodes.
In order to specify the time sources you need to take a sample chrony.conf file like below
server 0.rhel.pool.ntp.org iburst
server 1.rhel.pool.ntp.org iburst
server 2.rhel.pool.ntp.org iburst
server 3.rhel.pool.ntp.org iburst
driftfile /var/lib/chrony/drift
makestep 1.0 3
rtcsync
keyfile /etc/chrony.keys
leapsectz right/UTC
logdir /var/log/chrony
Modify your server lines to your custom NTP sources as needed then copy your contents into a URLencode tool.
Replace the line data line in the yaml file(s) on line 21 with
data:,URLENCODEDSRING......
Note: the data:, is critical to operation
With your custom configs you need to apply them to OpenShift. Soon as these machine configurations are applied the configs will be rendered to a new version of the master or worker machine config pools and rolling upgrades will start. If you wish you can apply one or both at the same time as a cluster-admin depending on the confidence.
To apply the config to the workers
oc apply -f 99-ntp-sources-worker.yaml
To apply the config to the masters
oc apply -f 99-ntp-sources-master.yaml
You should then see nodes cordon/disable scheduling and drain pods from their workloads
oc get nodes
And confirm it
NAME STATUS ROLES AGE VERSION
etcd-0.cluster.domain Ready master 2d21h v1.13.4+12ee15d4a
etcd-1.cluster.domain Ready master 2d21h v1.13.4+12ee15d4a
etcd-2.cluster.domain Ready master 2d21h v1.13.4+12ee15d4a
worker-0.cluster.domain Ready,SchedulingDisabled worker 2d21h v1.13.4+12ee15d4a
worker-1.cluster.domain Ready worker 2d21h v1.13.4+12ee15d4a
worker-2.cluster.domain Ready worker 2d21h v1.13.4+12ee15d4a
This process will continue for all the nodes in the cluster. You should be able to confirm their time sources after reboot by running the chronyc sources command via SSH.
Or as a part of one big command
for i in {etcd-0,etcd-1,etcd-2,worker-0,worker-1,worker-2}; do ssh -i cluster_rsa_key core@${i}.cluster.domain 'chronyc sources'; done
And you should see something like
210 Number of sources = 3
MS Name/IP address Stratum Poll Reach LastRx Last sample
===============================================================================
^* ntpsource1 1 7 377 128 -25us[ -33us] +/- 764us
^- ntpsource2 3 6 377 1 -25us[ -25us] +/- 7064us
^- ntpsource3 3 6 377 59 +112us[ +112us] +/- 6193us
^- ntpsource4 1 6 377 59 +34us[ +45us] +/- 834us